Research on industrial software training teaching system based on digital twin and intelligent digital human

Abstract

In response to issues in industrial software practical teaching such as disconnection from industry demands, insufficient interactivity, delayed feedback, and single-dimensional evaluation, this study explores the integrated application of digital twin and intelligent digital human technologies in engineering practical teaching. With the core objective of cultivating engineering talent, an intelligent digital human system for industrial software training is designed and developed based on the Unity engine, digital human modeling, and multimodal interaction technologies. The system adopts a multi-layer architecture of "hardware interaction layer—algorithm parsing layer—scene application layer," integrating four major modules: high-fidelity intelligent interactive digital humans, intelligent theoretical instruction, digital twin virtual practice, and intelligent skill assessment and evaluation, thus constructing a complete teaching chain of "theoretical instruction—practical training—skill evaluation." The system uses MetaHuman to achieve high-fidelity digital human modeling, employs BlendShape and a lip-sync library to realize speech-driven facial expressions, and restores real industrial scenarios through digital twin technology. Tests have verified its good interactivity, immersion, and teaching support effects, effectively enabling precise alignment of teaching and industry needs. This system provides a technical solution for the intelligent upgrade of industrial software training and offers a practical reference for the integrated application of digital twins and intelligent interaction in engineering education.